Assistance with broad distance bins and habitat heterogeneity

[E L]

Hello, 

I am trying to fit a distance model to a point count dataset, consisting of multispecies bird observations in distance bins of <30m, 30-100m, and >100m. I am an undergraduate student who is completely new to distance sampling.

So far, I have encountered the following issues:

1) The data do not follow the typical distribution expected of point count data

• Firstly, for all species, there is a strong spike in the <30m band (see images), probably because it is so broad that the lack of observations closer to the observer (e.g., between 0-10m) is not captured. This means that there is no gradual increase to a peak occurring at smaller distances. Would it help to use a line transect model instead, or create a 0m distance band to shift the distribution to the right?
• Secondly, for some species (see Distance_Model and Distance_Model_Magpie images), there are more observations in the >100m band than in the 30-100m band. I suspect that this is because my observations were conducted in farmland, so birds in a hedgerow across a field are more readily detected than birds within a field in the 30-100m band. How can I fit a distance model that takes this issue into account? Should I add a habitat covariate that indicates whether a bird was observed in a hedgerow or a field?

2) There are too few distance bins to fit a model with multiple parameters (e.g., a half normal key function + cosine adjustment model)

• This is a problem because I was hoping to fit a model with two peaks that would take into account the effects of birds in distant hedgerows.

I would greatly appreciate any advice or assistance in addressing these issues. It could be that my data are inappropriate to fit into a distance model - I would still be grateful to receive this confirmation from someone more knowledgeable than I.

[Eric Rexstad]

Esther

Welcome to the group. You are correct that the figures you provided showed a problem in fitting detection function models to the data as depicted. Were the data recorded in the field using the cutpoints you describe (<30, 30-100, >100)? If so, there is little that can be done to remedy the situation. Modelling point count data is difficult when recording exact distances; it is much more difficult with the broad (0-30m) first bin for the reason you describe. It is the shape of the fitted detection function at small distances that is central to producing sound estimates from point transect data. I don't think either of your suggested remedies will be of use.

Regarding your second point, point transect surveys assume availability of animals increases linearly as a function of distance from the point; i.e. the number of animals available to be detected increases as the area of the detection bands increases. Existence of hedgerows creates a discontinuity in this linear availability function. Adding a covariate to the detection function model will not solve this problem. The problem is reduced if there are a large number of replicate transects (~20), lessening the influence of a particular habitat feature on the data set. Non-linearity of the availability curve at large distances can be mitigated by truncation (see Buckland 2006), but with three distance bands your truncation options are extremely limited.

• Buckland, S. T. (2006). Point transect surveys for songbirds: Robust methodologies. The Auk, 123(2), 345–357. https://doi.org/10.1093/auk/123.2.345

As you note, your ability to fit parameter-rich models is impeded by the number of bins into which your data were collected. In any event, a multi-modal detection function is not the answer to the problem, because it is not detectability that is changing because of the hedgerow, instead it is the availability of animals that is changing.

If you have the opportunity to repeat the survey, collecting detections using exact distances enhances your modelling opportunities. Also having sufficient replicate transects can diminish the influence of some types of availability features.

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of E L <esther....@gmail.com>
Sent: 03 August 2025 15:57
To: distance-sampling <distance...@googlegroups.com>
Subject: [distance-sampling] Assistance with broad distance bins and habitat heterogeneity

 

--
You received this message because you are subscribed to the Google Groups "distance-sampling" group.
To unsubscribe from this group and stop receiving emails from it, send an email to distance-sampl...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/distance-sampling/234a62df-402f-43a2-916f-a94b2f1e6f8cn%40googlegroups.com.